C5a is important in the tubulointerstitial component of experimental immune complex glomerulonephritis


Dr Thomas R. Welch, State University of New York Upstate Medical University Department of Paediatrics 750 E. Adams St. Syracuse, NY, USA  E-mail: welcht@upstate.edu


Interstitial injury is the hallmark of glomerulonephritis which is progressing to end-stage renal disease (ESRD). In humans and experimental animals, we have shown that interstitial disease is accompanied by up-regulation of complement components in tubular epithelial cells. Glomerulonephritis was induced in mice by the intraperitoneal injection of horse spleen apoferritin (HSA) and lipopolysaccharide (LPS). In addition to wild-type C57/B6 mice, animals in which the C5a receptor had been deleted (C5aR KO) were used. Animals were killed after 3 or 6 weeks, and kidneys harvested. At three weeks, both groups had evidence of mild mesangial matrix expansion and increased cellularity; there were no crescents, sclerotic lesions, or interstitial disease. At six weeks, glomerular lesions were advanced, but identical in the two groups. Both groups had evidence of an identical pattern of C3 gene expression in the tubular epithelium by in situ hybridization. There was a marked difference, however, in the extent of interstitial injury. Wild-type animals had significantly greater numbers of infiltrating interstitial cells, greater expansion of the peritubular space, more tubular atrophy, and more apoptotic tubular cells than did C5aR KOs. The anaphylotoxic fragment of C5, C5a, is not likely to be important in the glomerular component of this model of progressive glomerulonephritis. On the other hand, the interstitial component is markedly attenuated in knockout animals. These data support a role for complement in the interstitial component of this glomerulonephritis model. They are consistent with our hypotheses of a role for complement in the progression of some forms of glomerulonephritis to ESRD.


The development of interstitial injury – cellular infiltration, tubular atrophy, peritubular oedema, and fibrosis – is the hallmark of glomerulonephritis that is evolving into end-stage renal disease [1]. This was considered a nonspecific finding. More recent work, however, suggests that the tubular injury per se plays an important role in deteriorating renal function [2].

A wide variety of glomerulonephritis models have been employed over the past several decades. These have provided much insight into the mechanisms of glomerular inflammation and injury. For the most part, however, they have not been used to address the mechanisms of tubulointerstitial injury.

Recently, investigations in rats have suggested that complement proteins, as a component of glomerular proteinuria, could mediate tubulointerstitial damage [3]. Earlier work had pointed to ammonia-mediated complement activation as a mediator of interstitial injury [4]. Although complement has long been studied as a mediator of glomerular injury, demonstration of a role for this system in the interstitial component of glomerulonephritis would be novel.

In addition to urine as a source of complement protein, we have shown that the proximal tubular epithelium itself can synthesize several components, both in humans [5–7] and in mice [8,9]. We have suggested that such local production and activation, perhaps up-regulated by glomerular cytokines gaining access to glomerular filtrate, might also play a role in the mediation of tubular injury. This work has been confirmed by other groups in normal kidneys [10], human glomerulonephritis [11], and experimental transplant rejection [12].

While information about the synthesis and localization of complement components in the interstitium is consistent with a pathogenic role, it does not establish complement dependence. To do this, it is necessary to manipulate the complement system in vivo, either by pharmacological depletion, blockade, or with the use of animals lacking a specific component or receptor. The potent anaphylatoxin, C5a, is released when C5 is cleaved by a classical or alternate pathway C5 convertase. In this study, we employed mice lacking the C5a receptor to test the hypothesis that C5 activation contributes to the interstitial component of a well-characterized experimental glomerulonephritis.



Wild type C57/B6 mice (WT) were obtained from the Jackson Laboratory (Bar Harbor, ME). Congenic animals in which the gene for the C5a receptor had been deleted by gene targeting (C5aRKO) were a gift from Dr Craig Gerard. These animals have been described previously [13]. This study was reviewed and approved by the animal use committee of the Children's Hospital Research Foundation.


Horse spleen apoferritin (HSA, Sigma, St. Louis, MO, USA) and lipopolysaccharide (LPS, Calbiochem, San Diego, CA, USA) were obtained as indicated.

Disease model

Pilot studies with the C5aRKO mice led to a modification of the murine glomerulonephritis model we previously described [8,9]. The treated animals received 4 mg HSA by intraperitoneal (IP) injection five days per week, and 0·05 mg LPS by IP injection once per week. The intent of these experiments was to compare a well-characterized disease [8,9] in WT and C5aRKO animals. However, we always include saline-treated (0.l ml of 0.15m NaC1 IP five times per week) animals in these studies to insure that no spontaneous disease is developing in the colony.

WT and C5aRKO mice were treated for three or six weeks. At these intervals, animals were placed in metabolic cages for urine collection, and then sacrificed by C02 narcosis. Blood was immediately obtained for biochemical analysis, and the kidneys were prepared as detailed. At each of the two time intervals, a control (saline-injected) animal (WT and C5aRKO) was sacrificed, along with 5–7 treated mice (WT and C5aRKO).

Tissue harvesting and processing

The kidneys were dissected free and sectioned. Tissue samples were treated with 10% formalin and embedded in paraffin for histology, or with 4% paraformaldehyde and then frozen in M-l embedding matrix (Shandon Lipshaw, Pittsburgh, PA, USA) for in situ hybridization.

Histological analysis

The cortex from a single complete haematoxylin and eosin-stained kidney section from each animal was scanned at 400× using a Sony DKC 5000 video system attached to a microscope and saved with Adobe Photoshop. Using the scanned images, the numbers of interstitial cells in each section were counted and expressed per mm2. The area of interstitial space on the sections was calculated on a MacIntosh G-3 computer using the NIH public domain image analysis software program. After exclusion of area occupied by tubular blood vessels and glomeruli, this area was then expressed as a percentage of the total area of the image scanned.

Finally, the number of tubules in cross-section displaying evidence of atrophy was calculated from a single complete kidney section stained with Periodic Acid-Schiff (PAS). For purposes of this analysis, a tubule was considered atrophic if it displayed circumferential PAS positive basement membrane thickening. While there may be earlier, more subtle changes in evolving tubular atrophy, this criterion assured uniform, unequivocal interpretation. Adjacent atrophic tubules (foci), when found, were counted as one.

Although the focus of this study was the tubulointerstitial lesion, glomeruli were also examined to assess the severity of the proliferative lesion. As an objective measure of the severity of the glomerular lesion, a complete cross section of a kidney from each animal was also examined, and the percentage of glomeruli with crescents was calculated. For this study, a crescent was defined as the presence of two or more layers of cells in Bowman's space.


The presence of apoptosis was determined by immunohistochemical analysis and quantification using the In Situ Cell Death Detection Kit-AP from Boehringer Mannheim-Roche (Indianapolis, IN, USA). Paraffin sections were used and processed according to the manufacturer's instructions. The entire cortical section from a slide from each animal was analysed at 100× and 400× and the number of apoptotic cells were expressed/mm2.

In situ hybridization

Paraformaldehyde fixed tissue was employed for C3 in situ hybridization, as previously described from our laboratory [6–9]. Sense strand riboprobes were included in all experiments as negative controls, and mouse liver was employed as a positive control. Sections were examined by dark field microscopy at 200× and 400×. The presence, distribution, and intensity of signal was recorded for each tissue, as we have described previously [6–9].


Glomerular lesions

There were no differences between WT and C5aRKOs in the appearance of the glomeruli. The glomerular lesions in these animals were similar to those described in our previous studies [8,9], and were generally uniform within the two time groups. At three weeks, glomeruli of treated mice had slight expansion of mesangial matrix and increase in mesangial cellularity. No crescents or sclerotic lesions were present.

By six weeks, glomerulonephritis was well advanced. Treated animals showed involvement of all glomeruli. Mesangial expansion was advanced, as was glomerular cellularity (Fig. 1a, b). Sclerotic areas were present in many glomeruli. The percentage of crescentic glomeruli did not vary between WT (20%) and C5aRKO (22%; P> 0·1), and there was no difference in the degree of mesangial matrix expansion or proliferation between the two groups. No glomerular lesions were evident at either time period in either type of saline-injected mice.

Figure 1.

(a–b) Representative glomeruli from (a) WT and (b) C5aRK0 mice, treated with HSA and LPS and killed at 6 weeks. Hematoxylin and eosin, 200×. (c–d) Interstitial areas from (c) WT and (d) C5aRKO mice, treated with HSA and LPS and killed at 6 weeks. Hematoxylin and eosin, 200×. (e) Detail of interstitial area from WT mouse, treated with HSA and LPS and killed at 6 weeks. Note atrophic proximal tubule (arrow). Periodic Acid Schiff, 200×. (f–g) TUNEL reaction for apoptosis on kidney tissue from (f) WT and (g) C5aRKO mice, treated with HSA and LPS and killed at 6 weeks. (alkaline phosphatase), 200×.

Interstitial cellularity

The mean numbers of interstitial cells present in the tissues of the animals are presented in Table 1, and the differences in the appearance of the interstitium at 6 weeks are illustrated in Fig. 1c,d At three weeks, the difference between the WT and C5aRKO mice was not significant. At six weeks, however, the mean interstitial cellularity was nearly double in the WT mice compared to the C5aRKOs. The interstitial cells were predominantly mononuclear in their histological morphology. Further definition of their phenotype was not done in these studies.

Table 1.  Histomorphometric measures in glomerulonephritis model
3 week6 week3 week6 week
  • *

    P < 0·001 compared to WT at same time;

  • P < 0·01 compared to WT at same time;

  • P < 0·05 compared to WT at same time;

  • §

    P > 0·1 compared to WT at same time.

Interstitial cells/mm2 (mean (SD))828 (68)1323 (169)691 (62)699 (177)*
Interstitial area percentage (mean (SD)) 12·5 (1·8) 21·4 (4·2) 8·6 (2·0) 12·8 (4·5)
Foci of tubular atrophy/section (mean (SD)) 1·2 (1·3)  7·0 (2·6) 0·6 (0·8)§ 1·0 (1·0)
Apoptotic tubular cells/mm2 (mean (SD)) 0·9 (0·7) 18·1 (13·3) 0·5 (0·7) § 5·2 (4·8)

Tubular atrophy

Tubular atrophy was rare at three weeks in the treated animals. By six weeks, however, tubular atrophy was prominent in the WT mice, but continued to be rare in the C5aRKOs (Fig. 1e; Table 1).

Peritubular space

Using the image analysis software, the area of the peritubular space (expressed as a percent of total cortical area, exclusive of glomeruli and vessels) was not significantly different between the WT and C5aRKO groups at three weeks. By six weeks, however, the WT mice showed significantly greater peritubular expansion (Fig. 1c,d; Table 1).


While some apoptotic tubular cells (expressed as cells per mm2) were observed at three weeks, there was no significant difference in numbers between wild-type and C5aRKO animals. At six weeks, however, C5aRKO mice had significantly fewer apoptotic tubular cells than did their WT counterparts (Fig. 1f, g; Table 1).

C3 gene expression

As we have shown previously, WT animals began to show occasional foci of C3 mRNA expression in proximal tubular epithelial cells at three weeks. This expression did not differ between WT and receptor knock out animals (Fig. 2a, b). At six weeks, both groups of animals displayed widespread C3 message, again with no difference between WT and C5aRKO mice (Fig. 2c, d). There was no difference in distribution of C3 message in these animals when compared to our previous studies using this model [8,9]. Specifically, there was no signal in glomeruli (except in the cells of Bowman's capsule), vessels, or infiltrating cells.

Figure 2.

In situ hybridization for C3 mRNA on WT (a, c) and C5aRKO (b, d) mice treated with HSA and LPS for 3 (a, b) or 6 (c, d) weeks before killing. 100×, dark field.

Clinical disease

All animals developed haematuria, proteinuria and elevated BUN levels by six weeks (Table 2). There were no significant differences in any of these between the two groups.

Table 2.  Clinical disease in glomerulonephritis model
3 week6 week3 week6 week
  • *

    P < 0·001 compared to wild type at the same time.

Hematuria (0–5+) (mean (SD)) 0.0 (0) 3·5 (1.6) 0·2 (0·4) 2·3 (1.8)
Proteinuria (0–4+) (mean (SD)) 2·3 (0·6) 3·0 (0) 2·0 (0·6) 3·7 (0·5)
BUN (mean (SD))27·0 (0)45·3 (17.6)19·3 (1.8)*33·5 (9.2)


The extent of interstitial involvement in glomerulonephritis is one of the best markers of the likelihood of progression to end-stage renal disease. Although once looked upon as a nonspecific finding, it is increasingly appreciated that interstitial injury contributes to deteriorating renal function in chronic glomerulonephritis [1,2].

The mechanisms by which a primary glomerular process can progress to involvement of the interstitium are poorly understood. Although a host of animal models of glomerulonephritis have been developed over the past three decades, these have almost uniformly been employed to study the glomerular lesions rather than the associated interstitial injury. Recently, however, two separate lines of investigation have suggested a potential role for the complement system.

In proteinuric glomerular disease, induced in rats by puromycin aminonucleoside (PAN), interstitial injury has been ascribed to tubular damage from proteins in the ultrafiltrate [3]. It has been hypothesized that filtered complement proteins may be important in this damage. In a recent study, support for this hypothesis was provided by demonstrating a protective effect of the complement regulatory protein Crry on proteinuric tubular injury in a rat PAN model [14]. Deposition of C5b-9 on tubular cells was shown in this study, and overexpression of Crry led to decreased deposition and milder tubular injury. Similar observations have been made in other model systems [15].

In a series of human and animal studies, we have similarly suggested a role for complement in tubular injury, but have posited an alternative process. By in situ hybridization and immunohistochemistry, we have shown that the major components of the alternative pathway convertase, C3 (6) and B (7), are synthesized in the human proximal tubular epithelium. These studies also demonstrated that tubular complement expression is restricted to areas of peritubular inflammation or interstitial injury. This, in turn, led to the hypothesis that such local complement synthesis acts as a mediator of the tubulointerstitial component of chronic glomerulonephritis. Inflammatory mediators accessing the tubular cells from the glomeruli were the presumed regulators of this tubular expression, a hypothesis consistent with existing in vitro studies.

The presence of complement components in the peritubular interstitium, of course, does not provide a mechanism for their activation. One potential for this was shown by Hostetter and Gordon in 1987. C3 in the ammonia-rich interstitium may be amidated to C3NH3[4,16]. C3NH3 can function as a C3 convertase, setting up a cascade which can ultimately result in ongoing activation and cleavage of C3 and C5.

Studies with human material are limited to single time points in the evolution of disease. Therefore, this hypothesis ultimately requires animal models for testing. Using a protocol similar to that employed in the present study, we have shown that proximal tubular C3 expression occurs in mice as well. By examining animals at weekly intervals, we were able to show that such expression followed the demonstration of IL-1 and IL-6 in the glomeruli, but preceded histological tubular injury [8]. This sequence was consistent with our hypothesis and was evidence against tubular complement expression being a nonspecific marker of cell injury. As was the case in human tissue [6], C3 expression in tubules was restricted to areas of tubular injury [8,9].

In the current series of experiments, we have addressed one of the effector arms of the complement system. Once the complement cascade has been triggered, these are a variety of ways in which tissue damage can result. Deposition of the membrane attack complex (MAC; C5b-9) on a cell, for example, can lead to lytic injury and cell death; sublytic amounts of MAC, rather than being directly lethal, may regulate the expression of a variety of genes by the affected cell [17].

Cleavage of C5 by the classical or alternative pathway C5 convertases is the first step in generating a MAC. This cleavage generates a small peptide fragment, C5a, which has a variety of independent effects mediated by its occupation of a widely distributed receptor [18]. C5a is a potent chemoattractant, mediating the ingress of inflammatory mediators of both neutrophilic and lymphocytic lineages [19]. On other cell types, it can lead to signal transduction, resulting in a variety of effects including apoptosis [20].

The availability of C5aRKO mice allowed us to dissect out specifically the contribution of this anaphylotoxic peptide. In at least one other model of inflammatory injury, these animals have displayed deficient chemotactic responses [13]. Mice deficient in C5 itself, in contrast, would be disabled both in the generation of C5a and in the formation of a MAC. Such mice, in a similar glomerulonephritis model, appear to have attenuation of both the glomerular and interstitial component of the disease [21]. This would support a MAC-related role for C5 in the glomeruli and a C5a related role in the interstitium.

The dramatic difference in interstitial cellularity between the WT and C5aRKO animals is consistent with C5a acting as a chemoattractant in the model. The TUNEL results further suggest that C5a mediated apoptosis may be an additional mediator of tubular damage. We have recently examined the expression of the C5a receptor in the human kidney. The receptor was identified by immunohistochemistry on the proximal tubules of normal kidney tissue as well as in proximal tubular cell lines. In contrast, there was minimal expression of the receptor in glomerular mesangial cells [22].

The time course of these findings is also consistent with our hypothesis of local complement synthesis contributing to tissue damage. Our previous work has shown that local C3 expression in this model does not become evident until about three weeks [8,9]. Consequently, one would not predict any noticeable difference in interstitial injury before this. In these experiments, significant differences in measures of interstitial damage were not present at three weeks, but were well established at six weeks.

Although the intent of this study was not specifically to study the glomerular lesion, it did not differ significantly between the WT and the C5aRKOs. This suggests that C5a is not as critical to the glomerular component of the lesion as it is to the tubular injury, despite the fact that mesangial cells also express C5aR. Further, this observation speaks against the difference in interstitial injury simply being a marker of variations in the severity of the glomerulonephritis itself.

The in situ hybridization data for C3 further support this hypothesis. Our previous work, both in humans and in WT C57/B6 mice, has demonstrated C3 expression in the proximal tubules of kidneys in which interstitial injury was present. Studies by others in cultured epithelial cells have shown that such cells release C3 through their basolateral surface [23], which for renal tubular cells would be into the interstitial space.

We have interpreted these observations as supporting the hypothesis that tubular complement expression promotes tubular injury, although an alternative explanation is also possible. An exuberant peritubular inflammatory response could lead to up-regulation of tubular C3 expression, in which case the tubular complement would be a result of the interstitial injury rather than a cause of it. The widespread proximal tubular C3 expression in C5aRKO animals lacking a significant interstitial lesion speaks against this alternative explanation.

The similar urine findings in the two groups (Table 2) is consistent with the absence of any significant differences in their glomerular lesions. Both groups had evidence of deteriorating renal function at six weeks. Although our hypothesis would indicate that the WT group would eventually have more severe or persistent renal insufficiency, the current data do not permit such a conclusion.

After decades of research on the role of the complement system in the mediation of glomerular injury, it is clear that there is now mounting evidence for an entirely distinct role in tubular injury.